Intentional looks facilitate faster responding in observers

Human interaction relies heavily on understanding others' mental states. Eye gaze, a key nonverbal communication tool, rapidly conveys social messages. While gaze-following is well-established, it's unclear whether this is due to directional cues alone or also to the communication of the gazer's mental state. This study uses three preregistered experiments to investigate this by manipulating the gazer's mental state (intentional vs. non-intentional gaze) and observing its effect on observer responses. The human eye's morphology, with high contrast between iris and sclera, facilitates quick gaze direction interpretation. While some research suggests gaze-following is simply a response to directional cues (similar to arrows), other studies demonstrate reduced gaze-following when the gazer's ability to see the target is questioned or when the gaze cue comes from a non-social agent like a robot. This study uses videos of real individuals making intentional and unintentional eye movements to examine whether the mental state information in the gaze cue impacts observers' perception and subsequent gaze-following response. The study's design is similar to previous work showing faster identification of endogenously chosen targets in reaching tasks, suggesting a possible parallel with gaze perception. The experiments involve observing videos of intentional (self-chosen) and unintentional (computer-instructed) gaze and measuring response times (RTs) in different tasks to determine if observers are sensitive to the intentionality information within the gaze cue.

The literature on gaze following is divided on whether it's primarily driven by directional information or mentalistic information. Studies supporting the directional account show similar attentional following for gaze and non-social cues like arrows. Conversely, studies supporting the mentalistic account show reduced gaze-following when the observer believes the gazer cannot see the target or when the gaze cue is from a non-social agent. Recent reviews highlight the contributions of both directional and mentalistic information to gaze signals. This study aims to clarify this debate by directly manipulating the mental state of the gazer and measuring its impact on observer behavior.

Three experiments were conducted using videos of two female gazers making left or right eye movements. In self-chosen trials, gazers freely chose their gaze direction; in computer-instructed trials, they were explicitly told where to look. Experiment 1: Observers (N=81) viewed videos paused before the gaze shift, reporting the upcoming gaze direction. Response times were compared between self-chosen and computer-instructed gaze. Experiment 2: Observers (N=73) viewed the same videos, but this time a peripheral target appeared at the gazed-at or non-gazed-at location. Response times were measured for congruent (target at gazed-at location) and incongruent (target at non-gazed-at location) trials, comparing self-chosen and computer-instructed gaze. Experiment 3: Observers (N=70) repeated Experiment 2 with varied cue-target onset times (0, 100, 300, 700 ms). The methodology included pre-registration of hypotheses and exclusion criteria (RT < 200 ms, RT > 1800 ms, <70% trials after exclusion). Data analysis involved repeated measures ANOVAs. An exploratory analysis using Optical Flow Analysis (OFA) examined eye region motion differences between self-chosen and computer-instructed gaze in the 2000 ms before gaze initiation.

Experiment 1: Observers responded significantly faster to self-chosen gaze compared to computer-instructed gaze (p=0.004). Experiment 2: A classic gaze-following effect was observed (faster responses to congruent targets), but no significant interaction between gaze-following and gaze intentionality. However, exploratory analysis revealed that response speed differences between self-chosen and computer-instructed gaze varied depending on the gazer, suggesting individual differences in communication effectiveness. Experiment 3: Replicated the gaze-following effect and the faster response to self-chosen gaze, especially at longer cue-target intervals (100, 300, and 700 ms). At 0 ms, self-chosen gaze was slower, potentially indicating different processing time scales for mentalizing and directionality. OFA revealed significantly more motion in the eye region before self-chosen gaze compared to computer-instructed gaze (p=0.020). This subtle motion might communicate mental states.

The findings support the idea that human gaze communicates both directional and mentalistic information. The faster responses to self-chosen gaze in Experiment 1, even before the gaze shift, suggest rapid, implicit processing of mental state information. The lack of interaction between gaze-following and gaze intentionality in Experiments 2 and 3 may indicate that these processes operate via different mechanisms on different time scales. Mentalizing might be rapid and implicit, while directionality computation is more deliberate and explicit. The individual differences in communication effectiveness highlight the importance of considering such variability in future research. The results suggest parallel processing, with rapid mental state attribution initially, followed by slower, more deliberate processing of gaze direction and meaning. The OFA analysis suggests a potential mechanism: subtle differences in eye motion during self-chosen vs. computer-instructed gaze, imperceptible to the naked eye, may drive the mental state perception.

This study demonstrates that human gaze conveys mental states, likely via subtle motion information in the eye region. Mental state information and directionality from gaze appear to be processed in parallel, with different time scales of influence. Future research could explore how deliberate processing influences these different aspects of gaze information, investigate the role of different sensory modalities in mental state perception, and use high-speed eye tracking to better understand the nature of the motion signal associated with mental state perception.

The study used only two gazers, limiting generalizability. While controlling for gender, future studies should investigate how intentionality perception is modulated by gazer and observer gender interactions. Real-life contextual factors may also influence mental state perception from gaze, and future research should consider these more complex situations.